We have updated our Terms of Service, Code of Conduct, and Addendum.

Looking for recommendations for a good services/processes config for a 1000+ edge node deployment?

Can anyone here recommend a good services/processes config for a 1000+ edge node deployment? We are running into an issue where edge nodes are pegging at 100% CPU after updating cribl to 4.0.3. I wonder if our leader is overwhelmed with requests from all of our edge nodes and if our leader configs could use some tweaks. We are running pretty beefy leader/worker nodes so CPU and memory are not an issue, I am wondering if its more of an IOPS thing. Thoughts?

Answers

  • If the edge nodes are the ones experiencing the high usage then I'm unsure why changes to the leader node would have any bearing.

  • Leader specs may still need modified just because of managing all those nodes depending on when the leader is most impacted by them.